Inhibitors of angiogenesis are an important addition to conventional chemotherapy. Among different "druggable" angiogenic factors, fibroblast growth factor-2 (FGF-2) is an attractive target for novel therapies because of its intricated involvement in tumor neovascularization, tumor cell proliferation and migration, and the acquisition of resistance to antiangiogenic therapies. FGF-2 bioavailability and activity is affected by several natural ligands, including the endogenous inhibitor of angiogenesis thrombospondin-1 (TSP-1). We hypothesized that the FGF-2-binding sequence of TSP-1 might serve as a template for the development of non-peptide inhibitors of angiogenesis. Computational biology and nuclear magnetic resonance spectroscopy approaches, major investigative tools in the characterizations of protein-protein interaction (PPI), were used to map the residues at the TSP-1/FGF-2 interface. The translation of this three-dimensional information into a pharmacophore model allowed screening a small molecule databases, identifying three FGF-2-binding, antiangiogenic small molecules, mimetic of TSP-1. Pharmacophore-based approaches are thus feasible tools to exploit naturally occurring PPI, by generating a set of lead compounds mimetic of endogenous proteins, as a starting point for the development of novel therapeutic agents.